Insect. Soc. 54 (2007) 34 – 41 0020-1812/07/010034-8 Insectes Sociaux DOI 10.1007/s00040-007-0909-x Birkhuser Verlag, Basel, 2007

Research article

Social and genetic structure of a supercolonial weaver , robsoni, with dimorphic queens

J.S. van Zweden1,2, M.E. Carew3,4, M.T. Henshaw1,5, S.K.A. Robson1 and R.H. Crozier1,3,*

1 School of Marine and Tropical Biology, James Cook University, Townsville, Queensland 4810, Australia, e-mails: [email protected], [email protected] 2 Present address: Department of Population Biology, University of Copenhagen, Universitetsparken 15, 2100, Copenhagen, Denmark, e-mail: [email protected] 3 School of Biochemistry, Chemistry and Genetics, La Trobe University, Bundoora, Victoria, Australia 4 Present address: CESAR, University of Melbourne, Parkville, Victoria 3052, Australia, e-mail: [email protected] 5 Present address: McKendree College, 701 College Road, Lebanon, Illinois 62254, USA, e-mail: [email protected]

Received 19 July 2006; revised 16 October 2006; accepted 25 October 2006. Published Online First 19 January 2007

Abstract. We studied a population of the Australian Introduction Polyrachis robsoni with regard to variation in the morphology of its winged queens using six newly- Morphological polymorphism between queens is wide- developed microsatellite markers. Morphometrically the spread in , with cases in at least seven subfamilies queens fell clearly into two groups, macrogynes and (Heinze and Keller, 2000). These cases range from simple microgynes, with the latter an isometric reduction of the size differences to both winged and wingless queens being former. Aggression tests showed that hostility between found in the same species. The interest in the cases of ants from different nests was minimal. Nests frequently simple size polymorphism stems from the fact that these contained numbers of both queen types, with microgynes may represent the initial stages of the process leading to about twice as numerous as macrogynes. Nestmate work- the winged/wingless dichotomy, to the presence of only ers, microgynes, and macrogynes, were significantly wingless queens if the winged ones are lost, or possibly to related to others within their caste, with macrogynes sympatric speciation. Typically, the smaller queens (mi- more highly related than the other castes. Relatedness crogynes) are isometric reductions of the larger queens values between these groups of nestmates were also (macrogynes), and together the two forms display a significant. Pairwise relatedness values were consistent bimodal frequency distribution of body size. Microgynes with both queen morphs producing workers. At the also typically have a reduced number of functional population level, microgynes from different nests were ovarioles compared to macrogynes, as well as reduced also significantly related and there was a weak inverse wings (Lachaud et al., 1999; Rppell et al., 2001a), relationship between pairwise relatedness value between although there are cases where the number of ovarioles is individuals and distance between nests. We conclude that the same between the two queen forms (Rppell et al., this species is supercolonial and that the two queen 1998). morphs are part of the same population. Explanations for the presence of dimorphic queens vary, in part because of the highly-variable life history Keywords: Polyrhachis weaver ants, microsatellites, di- stategies of the species in which they have been found. morphic queens, dispersal, relatedness, supercolonies. The presence and number of macrogynes and microgynes within individual colonies or populations has been variously described as reflecting a variety of colony founding strategies, genetic polymorphism, or frequency dependent selection, but it has been difficult to derive common overall patterns and explanations (Heinze and Keller, 2000). For example, microgyne numbers are * Corresponding author greatest in larger colonies of Ectatomma ruidum (La- . Soc. Vol. 54, 2007 Research article 35 chaud et al., 1999) and of Temnothorax rugatulus Methods (Rppell et al., 2001c). The general perception is that the larger size of Collection site and colony structure macrogynes reflects independent colony foundation producing monogynous nests, and the smaller microgynes We collected all 37 nests visible in a group of trees bordering the edges tend towards dependent colony foundation and are often of a small pond in the Botanical Gardens of Cairns, North Queensland, Australia (16830’’’S,145830’’’E). The pond runs SW to NE and is about associated with polygyny and limited gyne dispersal 100m wide; the collections were made on the NW and SE sides of the (Rppell and Heinze, 1999). The extent to which micro- terminus of the pond. The relative position of each nest was recorded, gynes act as intraspecific parasites and are adopted into along with the number of brood, workers and queens found in each nest. unrelated nests, or are adopted back into their natal To determine if different nests might belong to the same colonies, we conducted a series of laboratory based aggression tests as a bioassay colonies, remains unknown in most cases. But using for colony membership. A single individual from each of two test molecular markers to estimate relatedness between the colonies were placed in a circular observation chamber (diame- two queen types, Rppell and coworkers (2001b; 2003) ter=10 cm) and their first contact recorded on a aggression scale found for Temnothorax rugatulus that macrogynes pre- ranging from 0 to 4 (0=trophallaxis, 1=antennation, 2=touch without further interaction, 3=touch and avoidance, and 4=grabbing). This dominantly found colonies independently but micro- approach was replicated for a total of ten pairs of individuals from each gynes rejoin their natal nests. Although the same general pair of colonies tested and the average aggression scores used as a pattern is believed to occur in Temnothorax cf. andrei, measure of between-colony aggression. A total of 24 colonies (12 pairs queen morphology was not correlated with colony social of two) were selected from a variety of locations to provide a range of structure and a high rate of macrogynes rejoining natal between-colony distances, and individual colonies were only tested in one pair-wise comparison. colonies was suggested (Rppell et al., 2001a). The ant genus Polyrhachis is a taxonomically, ecolog- ically and socially diverse group. With over 450 species Queen morphology divided into 12 subgenera (Bolton, 1995; Dorow, 1995), nesting habits range from subterranean in either terres- For each individual queen we determined head length (from the posterior most point along the posterior occipital border to the anterior trial or intertidal habitats (Nielsen, 1997), to lignicolous, most point on the anterior clypeal border), head width (the distance lithophilic, and arboreal (Liefke et al., 1998; Robson, between the occipital lobes), mesosoma length (the first three 2004), with nests constructed of various combinations of postcephalic segments plus the propodeum, in side view) and meso- carton, larval and spider silk (Robson and Kohout, 2005). soma width (the distance between the two wing scars). Measurements were made using NIH Image 1.61 (National Institute of Health, USA) Colony structure can be highly variable, with some and images captured with a JVC digital video camera and a Leica MZ6 species establishing nests through either single or multi- binocular microscope. All measurements were taken three times from ple queens (Sasaki et al., 2005; Yamauchi et al., 1987) or each individual and the average was used in the analysis. obligatory parasitic relationships with other species of ants (Maschwitz et al., 2003). The Australian weaver ant Genetic analyses Polyrhachis robsoni (previously referred to as P. doddi? (Bellas and Hçlldobler, 1985), P. cf. doddi (Heinze and Microsatellites were isolated using methods described previously Hçlldobler, 1993), P. Cyrto 04 or P. Cyrto 05 (Kohout, (Crozier et al., 1999). 2000)) is a polydomous arboreal ant that constructs nests We used the microsatellite loci Prob5, Prob37, Prob90, Prob114, from the silk of its larvae (Kohout, 2006). Individual nests Prob130, and Prob220 to estimate relatedness values. All queens and at least six workers from each nest were analysed, to determine a measure can contain numerous dimorphic queens, each of which of allele frequencies of each nest and the population as a whole. has a full set of thoracic sclerites and seemingly functional DNA was extracted using a standard salt precipitation protocol wings. Both queen morphs appear capable of reproduc- (Miller et al., 1988; Strassmann et al., 1996). Briefly, we ground tion, possessing apparently functional ovaries, and to- individual thoraxes in 500 ml of grinding buffer (0.1M NaCl, 0.1M Tris- HCl (pH 9.1), 0.05M EDTA, 0.05% SDS) using a mini-pestle. Samples gether present a typical bimodal but continuous size were then incubated for 30 minutes in a 658C water bath, 86 mlof8M frequency distribution (Heinze and Hçlldobler, 1993). potassium acetate was then added and the mixture stored for 30–40 However, the sample sizes available to Heinze and minutes in 48C, to let salts and other minerals precipitate. The mixture Hçlldobler (1993) were relatively small and the actual was then centrifuged at 13000 rpm for 5 minutes, 400 ml of the reproductive output of individual queens was not deter- supernatant transferred to a new tube and combined with 1000 mlof À208C 100% High Grade Ethanol, and stored at À208C for at least 3 mined. In this study we use molecular markers for the first hours for the DNA to precipitate. Samples were then centrifuged at time for this species to infer relatedness patterns between 13000 rpm for 5 minutes, the ethanol poured off and the remaining macrogynes, microgynes and workers, examine the rela- pellet dried for 10 minutes in a Speed Vac. The DNA was re-suspended tionship between internest relatedness values and geo- in 50 ml of ultrapure ddH2O and then diluted six times for use in PCR. For PCR we used 96 well plates with a reaction volume of 10 mlin graphic distance, and clarify the queen size distribution. each well (5 ml genomic DNA, 3 ml ddH2O, 1 ml TaqPol-buffer, 0.4 ml 2.5 mM dNTP mix, 0.5 ml10mM primer mix, 0.1 ml5U/ml Taq Polymer- ase). Thermal cycling began with an initial denaturing step of 948C for 2 min. This was followed by 35 cycles of: denaturing at 948C for 30 sec, annealing at the locus-specific temperature for 30 sec and extension at 728C for 45 sec. A final extension for 4 min at 728C was used. The annealing temperatures, allele sizes and primers for each locus are given in Table 1. PCR products were visualized on 5% denaturing 36 J. S. van Zweden et al. Molecular ecology of a supercolonial weaver ant

Table 1. Microsatellite annealing temperatures, size range, and primer nucleotide sequences. Locus Size range Temp. Primer sequences Prob5 235–255 558C F: AGCGTTTTGTCGTTTTATTTC R: CCCCTTCCCGCCCTCG Prob37 204–228 558C F: CCGTCGTTTCTCTTTGGTCA R: CGTCCGTTGGTGTCTGTCAT Prob90 109–155 558C F: ATAACGACATTCATTCGAA R: TTTCTACTTTCAGCTTGGTC Prob114 191–241 538C F: GTCATCAGAGTTGGAGTTCCTT R: ATTCTATAAAAACACAATCAAATC Prob130 168–232 538C F: CACTATAAATTACTCGGCTGCG R: CGTATCCATTAGGTAATAAACTCG Prob220 166–242 558C F: TGTAACGAAAGGGAGAAAGG R: CGAACGCATAAAGCATGTC Clone sequences for the microsatellite loci have been deposited in GenBank under the accession numbers DQ836330–DQ836335

polyacrylamide gels using a Corbett Research GelScan 2000 running at standard methods [see ch. 14 of Sokal and Rohlf (1981)], yielding the 1200Vand 508C. One primer of each pair was end-labelled at the 5’ with regression coefficient, Pearsons correlation coefficient, and an either Hex (Prob5, Prob37) or Tet (the rest) fluorescent due. ANOVA table giving the significance level of the regression. Relatedness values within and between groups were estimated Summary statistics are presented as meanÆstandard deviation, using the methods of Queller and Goodnight (1989) as implemented in except for relatedness values, where 95% confidence limits are given. the program RELATEDNESS 5.0.8 (Goodnight and Queller, 1999).

We use the directionality in relatedness that bYX is the proportion of Ys genes identical to genes in X (Crozier and Pamilo, 1980, 1993). Average Results relatedness was estimated within and between the two queen morphs and workers, both within- and between nests. Calculations were not corrected for frequency bias, but population allele frequencies were Queen dimorphism calculated with nests weighted equally. The information from the nests sampled was used for the background population. Error estimates were As did Heinze and Hçlldobler (1993) we found all queen made by jackknifing over loci. Significance for the relationship between measurements were distributed bimodally. Microgynes inter-nest symmetric relatedness and inter-nest distance was estimated using Mantel tests implemented using GenAlEx (Peakall and Smouse, are an isometric reduction of macrogynes (Fig. 1). The 2006). For paired tests of relatedness values between classes of nests measurements were correlated (Pearson correlation co- (e.g., monogynous versus polygynous nests) we used the test within efficient, p<0.01). Considering all four measurements, RELATEDNESS. queens could be classified unambiguously as either To test whether or not one pair of nests with high worker-worker relatedness values could actually have been drawn from a common macrogyne or microgyne. Mesosoma lengths (ML) of distribution with another class of nests, we used a randomization to test microgynes ranged from 2.15 to 2.71 mm to see if a pair of nests with such a high mean value could have been (meanÆS.D. =2.42Æ1.4 mm) and did not overlap with drawn by chance. To do this we made a common distribution of the those of macrogynes, which ranged from 2.77 to 3.30 mm individual intra-nest relatedness values and drew two nests without = < replacement from this distribution, and repeated the procedure in all (3.01 Æ1.3 mm, ANOVA F 1,94 388.2 p 0.001). Simi- 10,000 times. The mean of the relatedness values of the two actual nests larly, the mesosoma widths of microgynes ranges from was then compared with this distribution. Because this is an a posteriori 1.17 to 1.66 (150Æ0.08 mm) and did not overlap with test, it is a two-tailed one and we rejected the two nests being part of a those of macrogynes, which ranged from 1.72 to 2.05 common group if their mean relatedness value fell more than 97.5% (1.88 Æ0.09 mm, ANOVA F =406.5, p<0.001). Con- along the upper end of the distribution (or lower than 2.5% at the lower 1,94 end). sidering both mesosoma length and mesoma widths, the Correction for multiple comparisons in tables of significance test two morphs fall into two distinct and non-overlapping results were made using QVALUE (Storey, 2003; Storey and Tibshir- groups. A value of 4.4 for the composite variable ani, 2003), which uses an approach which is an extension of the false ML+MW discriminates between the two morphs. discovery rate (Benjamini et al., 2001). QVALUE converts a set of p values to q values, the latter being calculated in the context of the full The number of queens, the number of queen morphs, distribution of p values such that, when the significance rate is set at 5%, and the number of workers found in individual nests were the set of q values which are listed as significant include no more than highly variable. Twelve of the 37 nests collected were 5% tests which are actually null results. queenless and six had one queen each. Of the 18 In order to determine if individuals move between nests we inferred the relatedness values between individuals using the program polygynous nests, four contained only microgynes and KINSHIP 1.3.1 (Goodnight and Queller, 1999). We tested for the two only macrogynes. Thirty of the 96 queens collected relationships Mother-Daughter versus unrelated, Sister-Sister versus were macrogynes and 66 were microgynes. Queen unrelated, Mother-Daughter versus Sister and Sister versus Mother- number per nest ranged from 0 to 16 (3.1Æ3.9), with Daughter. Because the number of tests is large there is a multiple the number of macrogynes ranging from 0 to 7 (1.0Æ1.5), comparisons problem under which some false positive results are expected when the number of tests is large. The q value approach that of microgynes from 0 to 13 (2.0Æ3.1) and that of requires exact calculation of p values; RELATEDNESS and KINSHIP workers from 2 to 115 (44.4Æ33.4). do not supply exact p values, so that q values cannot be determined for For nests in which they were present, the number of their outputs. We estimated the 95% threshold numbers of tests for microgynes per nest was positively correlated with the these numbers of tests at the 5% significance level using the normal distribution and simulations. amount of brood (Spearman rS =0.65, t30 =4.68, We used the package STATVIEW 4.5 (Abacus Concepts, Berkeley, p<0.001) and the number of workers ( rS =0.60, CA) to calculate simple regression for the morphometrics, using t30 =4.14, p<0.001). Insect. Soc. Vol. 54, 2007 Research article 37

Figure 2. Relatednesses between and within groups within nests. Only relatednesses significantly higher than zero are shown, together with their 95% confidence limits. Nests were weighted equally and jack- knifing was over loci. Figure 1. Plot of queen mesosoma length versus mesosoma width in mm, showing two distinct groups, the microgynes and macrogynes. the monogynous nests, with the mean of their related- nesses falling at result number 9237 of the 10,000 The distance between the pairs of nests in this study simulated results. We therefore combined the queenless ranged from 0.1m to 248m (45.5Æ77.7 m), but the level of and monogynous nests and found that the relatedness aggression between workers from different nests was value of this group did not differ from that of the independent of the distance between them (Spearman polygynous nests (Table 2). rS =À0.37, t10 =0.29). Mean levels of aggression between nests ranged from 0.2 to 1.1 (0.7Æ0.28, n=10), indicating that aggression was negligible. Table 2. Within nest relatedness comparisons between colonies of different social structure Colony type (n) relatedness C.L. Group Relatedness patterns: overall Queen morphs Macrogynes only (3) 0.2143 0.1361 a Ants within nests were related (Fig. 2). The highest relatedness was between macrogynes, but there was also a Microgynes only (5) 0.0942 0.0858 a high level of relatedness of these to nestmate microgynes, Mixed colonies (12) 0.1099 0.0795 a and vice versa. Worker nestmates were significantly Queen numbers related to each other and to the queen morphs, but less Polygynous (18) 0.1494 0.0687 a than the queen morphs were related to each other. Monogynous (6) 0.2146 0.0780 a Population-wide, macrogynes from different nests Queenless (2) 0.3960 0.1037 b were not related to each other, and neither were workers, but microgynes were (0.11 Æ0.07, CI). Restricting the Queen numbers, assuming queenless nests were monogynous calculation to only nests containing microgynes reduced Polygynous (18) 0.1494 0.0687 a the level of relatedness, but it remained significant Monogynous (8) 0.2523 0.0733 a (0.09 Æ0.06, CI). C.L.=95% confidence intervals, n=number of colonies of type shown, Tests of relatedness values within nests according to shared Group letter within comparison type indicates relatedness val- class of queen present found that all three groups (only ues not significantly different within class of test (morph or number). macrogynes present, only microgynes, or both kinds of queen) had significant relatedness values, but that these values were not significantly different between nest We could not exclude the possibility that at least some classes (Table 2). Nests classed by queen number (polygy- of the apparently monogynous nests had in fact been nous,ACHTUNGRE monogynous, or queenless) also showed significant polygynous but one or more additional queens had died, relatedness values for all classes, but the two queenless departed before collection, or been missed. If this was the nests had a significantly higher value than the other two case, then the difference in relatedness between monog- classes (Table 2). However, there was the possibility that ynous and polygynous nests might have been greater than the queenless nests were or had been monogynous. A we estimated. randomization test showed that the two nests could have been drawn by chance from a group comprising them and 38 J. S. van Zweden et al. Molecular ecology of a supercolonial weaver ant

Relatedness patterns with distance as mothers of workers. Highly significant relatedness values were also found between some queens in different Symmetric relatedness values between workers, micro- nests. gynes, and macrogynes, between pairs of nests plotted Relatedness patterns are consistent with microgynes against the distance between nests are shown in Figure 3. producing workers, though the evidence is weak. KIN- As shown in Table 3, there was a significant decline of SHIP results excluded sister-sister relationships for 22 out inter-nest relatedness with distance for all three classes of of 45 microgyne-worker pairs in favor of mother-daughter ant. Where the numbers of nests was large enough to ones, but could not exclude one or other of cousin-cousin permit this, the relationship was also assessed for the nests or aunt-niece (or both) for these pairs. on either side of the pond separately. There was a general picture of a significant decline of inter-nest relatedness value with distance. Discussion

The results point to the both queen morphs being of the Table 3. Mantel test results of the relationship between inter-nest relatedness and distance. same population and hence conspecific. Both morphs are significantly related to all three nestmate categories, and Group Number of nests b P both may yield workers, although the evidence on fertilities is not conclusive. The dimorphism in Polyrha- All workers 26 À0.0011 0.001** chis robsoni is thus similar to other ant species with NW workers 16 À0.0013 0.069 dimorphic queens such as Myrmica rubra (Steiner et al., SW workers 10 À0.0061 0.001** 2006), Myrmica ruginodis (Elms cited by Rppell and Heinze (1999), Temnothorax cf. andrei (Rppell et al., All microgynes 17 À0.0012 0.001** 2001a), T. rugatulus (Rppell et al., 2001b), and Ecta- NW microgynes 14 +0.003 0.046 tomma ruidum (Lachaud et al., 1999), and is not a case of All macrogynes 15 À0.002 0.001** separate reproductive communities as in E. tuberculatum NW macrogynes 11 À0.0075 0.007** (Hora et al., 2005). Polyrhachis robsoni differs from some species in that Significance levels <0.05* and <0.01** following Q-value adjustment are indicated. Cases where fewer than five nests could be included were there was no tendency of macrogynes to occur in not tested. monogynous nests; in fact nests sometimes held numbers of both forms. However, the increasing proportion of queens which were microgynes with increasing colony Relatedness patterns: pairwise relationships size reflects the situation in various other species. The lower relatedness of microgyne than macrogyne nest- The complete set of 238 individuals genotyped yielded mates might stem from a variety of causes, such as a 28,202 tests (one pairing could not be tested due to greater tendency of microgynes to return to the natal nest, missing values). The threshold for 95% significance out of or to another, or a shorter life span of the smaller queens. this number is 1470. A shorter life span would favor there being more 2073 pairings were significant for the Sister versus microgyne than macrogyne generations during the life Unrelated test, of which 194 tests pertained to within-nest of a nest. comparisons and 1879 to between-nest comparisons. The low levels of hostility between workers from These same pairings were also significant for the Moth- different nests, the significant relatedness values between er-Daughter versus Unrelated test. Thus, many of the individuals in different nests, and the decline in related- individuals in different nests were closely related, even if ness value with distance, are consistent with this popula- we cannot be sure which ones they were. tion being a single supercolony. Supercoloniality is not Testing the 2073 significant results (for the Sister clearly associated with winged queen dimorphism; for versus Unrelated test) for the Sister Versus Mother- example, while Pseudomyrmex venifica has supercolonies Daughter test yielded 96 significant results, but this (Janzen, 1973) Temnothorax cf. andrei and T. rugatulus do number is less than the 95% threshold number of 119. not, although colonies re-adopt microgyne daughters However, 1051 significant tests resulted from the Mother- (Rppell et al., 2001a,b). The Polyrhachis robsoni case Daughter versus Sister test. Of these, 88 were within-nest clearly resembles cases such as that of unicolonial net- tests and 963 between-nest tests. Hence a large number of works in Formica species, often highly viscous genetically individuals in different nests are related as mother to (Chapuisat et al., 1997; Pamilo et al., 1997, 2005), rather daughter, even though we cannot be sure which ones than of Linepithema humile, characterised by large-scale these are. movement within supercolonies leading to genetic ho- Significant relatedness values within nests were seen mogenization (Jaquiry et al., 2005). between some workers as well as between workers and Heinze and Hçlldobler (1993) made the attractive queens and between queens and queens. Both macro- suggestion for P. robsoni that macrogynes are capable of gynes and microgynes were identified with high certainty flight and independent colony formation whereas the Insect. Soc. Vol. 54, 2007 Research article 39

Figure 3. Symmetric internest relatedness versus internest distance. See Methods for a site description and Table 3 for the significance of the relationships assessed using Mantel tests. 40 J. S. van Zweden et al. Molecular ecology of a supercolonial weaver ant microgynes are not. The relatedness patterns do not Crozier R.H. and Pamilo P. 1980. Asymmetry in relatedness: who is reflect this concept, in that relatedness value declines with related to whom? Nature 283: 604 distance over a fairly short interval for both queen forms. Crozier R.H. and Pamilo P. 1993. Sex allocation in social : problems in prediction and estimation. In: Evolution and Diversity However the suggested role in colony formation is not of Sex Ratio in Insects and Mites (Wrensch D.L. and Ebbert M.A., excluded. New supercolonies may be founded exclusively Eds), Chapman & Hall, New York. pp 369–383 by macrogynes and subsequent movement within these Dorow W.H.O. 1995. Revision of the ant genus Polyrhachis Smith, 1857 supercolonies then occurs by walking. (: Formicidae: ) on subgenus level with It is reasonable to adapt Maynard Smiths (1964) keys, checklist of species and bibliography. Cour. Forsch.-Inst. Senckenberg 185: 1–113 haystack model to cases of supercoloniality (see also Goodnight K.F. and Queller D.C. 1999. Computer software for Crozier (1977)). In this model, a population once founded performing likelihood tests of pedigree relationship using genetic survives for a few generations before expiring. In the markers. Mol. Ecol. 8: 1231–1234 present case, and following Heinze and Hçlldobler (1993) Hamilton W.D. and May R.M. 1977. Dispersal in stable habitats. Nature 269: 578–581 in assuming that macrogynes are capable of independent Heinze J. and Hçlldobler B. 1993. Queen polymorphism in an colony foundation and microgynes are not, selection Australian weaver ant, Polyrhachis cf. doddi. Psyche 100:83–92 within populations may favor microgynes, in that a Heinze J. and Keller L. 2000. Alternative reproductive strategies: a developing larva needs fewer resources to metamorphose queen perspective in ants. Trends Ecol. Evol. 15: 508–512 Hora R.R., Doums C., Poteaux C., Feneron R., ValenzuelaJ., Heinze J. as a queen, but the effects of population extinction are to and Fresneau D. 2005. Small queens in the ant Ectatomma favor macrogynes. As Rppell et al. (2001c) note, as a tuberculatum: a new case of social parasitism. Behav. Ecol. supercolony expands and saturates the habitat, readop- Sociobiol. 59: 285–292 tion becomes favored and likely to select for microgyny. If Janzen D.H. 1973. Evolution of polygynous obligate acacia-ants in supercolonies are sufficiently long lived, macrogynes may western Mexico. J. Anim. Ecol. 42: 727–750 Jaquiry J., Vogel V. and Keller L. 2005. Multilevel genetic analyses of be lost. Those ants with queens with obligatorily depend- two European supercolonies of the Argentine ant, Linepithema ent colony foundation (Boomsma et al., 1995) may humile. Mol. Ecol. 14: 589–598 represent the end point of such a trend, although this Kohout R.J. 2000. A review of the distribution of the Polyrhachis and poses a continuing puzzle because even in stable con- Echinopla ants of the Queensland wet tropics (Hymenoptera: formicidae: Formicinae). Mem. Qld. Mus. 46: 183–209 tinuing habitats selection is expected to favor a residual Kohout R.J. 2006. Review of Polyrhachis (Cyrtomyrma) Forel (Hyme- level of dispersal (Hamilton and May, 1977). But species noptera: Formicidae: Formicinae) of Australia, Borneo, New with dependent colony foundation do disperse, although Guinea and the Solomon Islands with descriptions of new species. not as far as those with winged independent queens. Mem. Old. Mus. 52: 87–146. Lachaud J.P., Cadena A., Schatz B., Perez-Lachaud G. and Ibarra-Nunez G. 1999. Queen dimorphism and reproductive capacity in the ponerine ant, Ectatomma ruidum Roger. Oecologia 120: 515 – 523 Acknowledgements Liefke C., Dorow W.H.O., Hçlldobler B. and Maschwitz U. 1998. Nesting and food resources of syntopic species of the ant genus Polyrhachis (Hymenoptera, Formicidae) in West-Malaysia. Insect. JSvZ visited Australia under a study-abroad program of the University Soc. 45: 411–425 of Amsterdam with financial support from the AUV Fund (University Maschwitz U., Go C., Dorow W.H.O., Buschinger A. and Kohout R.J. of Amsterdam), the Dr. H. M. Vaderlandsch Fund and the Bekker-La 2003. Polyrhachis loweryi (Formicinae): A guest ant parasitizing Bastide Fund. This work was supported by grants from the Australian Rhytidoponera sp. (Ponerinae) in Queensland, Australia. Insect. Research Council to RHC and SKR (DP0209447, A00000732). We Soc. 50:69–76 thank Ching Crozier for advice in the laboratory, and two anonymous Maynard Smith J. 1964. Group selection and kin selection. Nature 201: referees for helpful comments. 1145–1147 Miller S.A., Dykes D.D. and Polesky H.F. 1988. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 16: 1215–1215 References Nielsen M.G. 1997. Nesting biology of the mangrove mud-nesting ant Forel (Hymenoptera, Formicidae) in north- Bellas T. and Hçlldobler B. 1985. Constituents of mandibular and ern Australia. Insect. Soc. 44:15–21 Dufours glands of an Australian Polyrhachis weaver ant. J. Chem. Pamilo P., Gertsch P., Thorn P.and Sepp P.1997. Molecular population Ecol. 11: 525–538 genetics of social insects. Annu. Rev. Ecol. Syst. 28:1–25 Benjamini Y., Drai D., Elmer G., Kafkafi N. and Golani I. 2001. Pamilo P., Zhu D.Q., Fortelius W., Rosengren R., Sepp P. and Controlling the false discovery rate in behavior genetics research. Sundstrçm L. 2005. Genetic patchwork of network-building wood Behav. Brain Res. 125: 279–284. ant populations. Ann. Zool. Fenn. 42: 179–187 Bolton B. 1995. A New General Catalogue of the Ants of the World. Peakall R. and Smouse P.E. 2006. GENALEX 6: genetic analysis in Harvard University Press, Cambridge, Massachusetts. 504 pp Excel. Population genetic software for teaching and research. Mol. Boomsma J.J., Keller L. and Nielsen M.G. 1995. A comparative analysis Ecol. Notes 6: 288–295 of sex ratio investment parameters in ants. Funct. Ecol. 9: 743–753 Queller D.C. and Goodnight K.F. 1989. Estimating relatedness using Chapuisat M., Goudet J. and Keller L. 1997. Microsatellites reveal high genetic markers. Evolution 43: 258–275 population viscosity and limited dispersal in the ant Formica Robson S.K. 2004. Comparative nesting biology of two species of paralugubris. Evolution 51: 475–482 Australian lithocolous ants: Polyrhachis (Hedomyrma) turneri Crozier R.H. 1977. Evolutionary genetics of the Hymenoptera. Annu. Forel and P.(Hagiomyrma) thusnelda Forel (Hymenoptera : Rev. Entomol. 22: 263–288 Formicidae : Formicinae). Aust. J. Entomol. 43: 5–9 Crozier R.H., Kaufmann B.E., Carew M.E. and Crozier Y.C. 1999. Robson S.K.A. and Kohout R.J. 2005. Evolution of nest-weaving Mutability of microsatellites developed for the ant Camponotus behaviour in arboreal nesting ants of the genus Polyrhachis Fr. consobrinus. Mol. Ecol. 8: 271–276 Smith (Hymenoptera : Formicidae). Aust. J. Entomol. 44: 164–169 Insect. Soc. Vol. 54, 2007 Research article 41

Rppell O. and Heinze J. 1999. Alternative reproductive tactics in Sokal R.R. and Rohlf F.J. 1981. Biometry, 2nd edition. Freeman, San females: the case of size polymorphism in ant queens. Insect. Soc. Francisco. 859 pp 46: 6–17 Steiner F.M., Schlick-Steiner B.C., Konrad H., Moder K., Christian E., Rppell O., Heinze J. and Hçlldobler B. 1998. Size-dimorphism in the Seifert B., Crozier R.H., Stauffer C. and Buschinger A. 2006. No queens of the North American ant Leptothorax rugatulus (Emery). sympatric speciation here: multiple data sources show that the ant Insect. Soc. 45:67–77 Myrmica microrubra is not a separate species but an alternate Rppell O., Heinze J. and Holldobler B. 2001a. Genetic and social reproductive morph of Myrmica rubra. J. Evol. Biol. 19: 777–787 structure of the queen size dimorphic ant Leptothorax cf. andrei. Storey J.D. 2003. The positive false discovery rate: A Bayesian Ecol. Entomol. 26:76–82 interpretation and the q-value. Ann. Stat. 31: 2013–2035 Rppell O., Heinze J. and Hçlldobler B. 2001b. Alternative reproduc- Storey J.D. and Tibshirani R. 2003. Statistical significance for genome- tive tactics in the queen-size-dimorphic ant Leptothorax rugatulus wide studies. Proc. Natl. Acad. Sci. U. S. A. 100: 9440–9445 (Emery) and their consequences for genetic population structure. Strassmann J.E., Solis C.R., Peters J.M. and Queller D.C. 1996. Behav. Ecol. Sociobiol. 50: 189–197 Strategies for finding and using highly polymorphic DNA micro- Rppell O., Heinze J. and Hçlldobler B. 2001c. Complex determination satellite loci for studies of genetic relatedness and pedigrees. of queen body size in the queen size dimorphic ant Leptothorax In:Molecular Zoology: Advances, Strategies, and Protocols (Fer- rugatulus (Formicidae : Hymenoptera). Heredity 87:33–40 raris J.D. and Palumbi S.R., Eds), Wiley-Liss, Inc. pp 163–180, 528– Rppell O., Stratz M., Baier B. and Heinze J. 2003. Mitochondrial 549 markers in the ant Leptothorax rugatulus reveal the population Yamauchi K., It Y., Kinomura K. and Takamine H. 1987. Polycalic genetic consequences of female philopatry at different hierarchical colonies of the weaver ant Polyrhachis dives. Kontyu 55: 410–420 levels. Mol. Ecol. 12: 795–801 Sasaki K., Jibiki E., Satoh T. and Obara Y. 2005. Queen phenotype and behaviour during cooperative colony founding in Polyrhachis moesta. Insect. Soc. 52:19–25